Bioacoustics of Eumops perotis

In February 2021 , fourteen Western mastiff bat echolocation passes containing 30 high-quality were recorded in the Arica and Parinacota region , northern Chile , during a rapid bat diversity assessment. The locality records were the valleys of Azapa ( 18°30′51.3″S 70°12′19.84″W ), Camarones ( 19°0′39.04″S 69°48′43.83″W ), and Lluta ( 18°24′38.0″S 70°08′ 01.85″W ) ( Figure 1 ). Bat passes were recorded using an M500- 384 USB Ultrasound Microphone ( Wildlife Acoustics , Inc. , Maynard, MA , USA), coupled to a Smartphone Samsung Galaxy S 20 FE, and were analyzed using BatSound 2.1 software ( Pettersson Elektronic AB , Upsala , Sweden ). Recordings were conducted using a sampling frequency of 384 kHz. The habitat where echolocation passes were recorded included thorny forests and scrubland, composed primarily of Geoffroea decorticans and Prosopis alba , surrounded by large areas with no vegetation ( Luebert and Pliscoff 2017 ) and were at medium elevations (approximately 223–800 m a.s.l). In particular, the Azapa and Lluta valleys are strongly influenced by human activity, mostly by agriculture (crops of alfalfa, corn, mango, guava, and olive tree plantations) and human settlements. Jaime A. Pacheco, Centro de Investigación Biodiversidad Sostenible (BioS), Lima , Perú , E-mail: biojpacheco@gmail.com Clemente Beltrán and Juan Luis Allendes, Bioecos E.I.R.L,Programa para la Conservación de los Murciélagos de Chile (PCMCh), Santiago , Chile , E-mail: cabeltran@uc.cl (C. Beltrán), jrallend@gmail.com (J.L. Allendes) Joaquín A. Ugarte-Núñez, Asociación para la Conservación y el Desarrollo Sostenible Sallqa Perú , Arequipa , Perú , E-mail: joaquin.ugarte@sallqaperu.org Each pulse was manually identified and assigned to E. perotis using visual identification after comparing the parameters of our recorded calls with reference calls of this species from MartÍnez–Gallardo et al. (2008) , Arias-Aguilar et al. (2018) , and Ortega et al. (2022) . Even though many molossid bats possess a high plasticity in echolocation call design, pulses of E. perotis are easily recognizable from other bat species because of their lower frequency range ca. 14.2 to 8.6 kHz (audible to the human ear) and longer call duration (>20 ms) (see Arias-Aguilar et al. 2018 and Ortega et al. 2022 for a comparison among species). These spectral and temporal features are diagnostic for this species allowing easy recognition ( Pierson 1998 ). A Discriminant Function Analysis (DFA) was also used to predict the probability of our record echolocation calls of belonging to E. perotis by comparing the parameters of our calls to a library of validated reference calls of molossid bat species with similar call designs occurring in our study area and surroundings. The library consisted of 429 full-spectrum recordings from free-flying bats of Tadarida brasiliensis ( n = 42 pulses), Nyctinomops aurispinosus ( n = 123 pulses), Mormopterus kalinowskii ( n = 204 pulses), E. perotis ( n = 30 pulses) and E. chiribaya ( n = 30 pulses). This last species is the only one in the genus that overlaps in distribution with E. perotis in southern Peru and could potentially occur in Chile ( Ugarte-Núñez 2020 ). Although it also occurs in our study area, we did not include calls of Promops davisoni since their structure (pulses that increase or decrease in frequency) differs markedly from those of other genera such as Tadarida , Eumops , Nyctinomops , and Mormopterus ( Jung et al. 2014 ; Ossa et al. 2018 ; Ugarte-Núñez 2020 ). Echolocation calls from the library were used as the known groups to calculate the discriminant functions used to predict which species our recorded calls belong to. Variables used in the DFA were call duration, end and start frequencies, slope frequency modulation, peak frequency, and interpulse interval. These variables were standardized to a mean of zero and a standard deviation of one, before computing DFA to make their scale comparable. λ -Wilks’ values were obtained with an MANOVA to test for the statistical significance of DFA models. Analyses were conducted in the software RStudio (version 2022.02.1 ) using MASS , dplyr, and ggplot2 packages. Echolocation calls of E. perotis in our study are characterized by pulses of long duration (mean ± SE = 17.06 ± 0.68 ms; range = 25.00–12.20) emitted at long intervals (879 ± 85.78 ms; 2580–415). Each pulse consists of a quasi-constant frequency component starting at 14.73 ± 0.35 kHz (21.3–12.1) and ending at 10.30 ± 0.18 kHz (12.5–8.9), with a peak frequency of 12.73 ± 0.18 kHz (15.2–11.1), and low slope frequency modulation (0.28 ± 0.02 kHz/ms; 0.52–0.12) ( Figure 2 ). These spectral and temporal features are consistent with those reported for specimens of the species from Peru , Mexico and Brazil ( Arias-Aguilar et al. 2018 ; MartÍnez– Gallardo et al. 2008 ; Ortega et al. 2022 ; Ugarte-Núñez 2020 ) supporting the identity of our recordings. Multivariate DFA gave an overall classification of 97.2% of the pulses (Wilk’ s λ = 0.0086, χ 2 = 2152.2, p <0.000), with the first discriminant function accounting for 90.9% of the total variation. Our prediction model correctly classified 86.6% of the recorded echolocation calls assigning them to E. perotis ( Figure 3 ). The lower frequency ranges (15–10 kHz) and longer call duration (>15 ms), set our echolocation calls apart from the other molossid bat species occurring in southern Peru and Chile , giving us a high level of confidence that we assigned these pulses to the correct species.